CN111612243A - Traffic speed prediction method, system and storage medium - Google Patents
Traffic speed prediction method, system and storage medium Download PDFInfo
- Publication number
- CN111612243A CN111612243A CN202010418444.2A CN202010418444A CN111612243A CN 111612243 A CN111612243 A CN 111612243A CN 202010418444 A CN202010418444 A CN 202010418444A CN 111612243 A CN111612243 A CN 111612243A
- Authority
- CN
- China
- Prior art keywords
- graph
- matrix
- traffic speed
- new
- traffic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 66
- 239000011159 matrix material Substances 0.000 claims abstract description 85
- 238000012549 training Methods 0.000 claims abstract description 31
- 238000012360 testing method Methods 0.000 claims abstract description 19
- 238000013528 artificial neural network Methods 0.000 claims description 28
- 238000013527 convolutional neural network Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 17
- 239000013598 vector Substances 0.000 claims description 14
- 238000001228 spectrum Methods 0.000 claims description 13
- 230000002123 temporal effect Effects 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 6
- 238000005065 mining Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000014509 gene expression Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 8
- 230000000694 effects Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 238000013135 deep learning Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000015654 memory Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 239000000523 sample Substances 0.000 description 3
- 206010039203 Road traffic accident Diseases 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Human Resources & Organizations (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a traffic speed prediction method, a system and a storage medium, wherein an original traffic speed data set is collected, the data set is divided into a training set and a test set, different spatial relations of the original traffic speed data set are mined, and two road network graphs are constructed; fusing the adjacency matrixes of the two road network graphs into a new graph adjacency matrix; taking the training set and the new image adjacency matrix as the input of a traffic prediction model, and training to obtain a prediction model; and predicting the traffic speed by using the prediction model. The method can fully and comprehensively excavate the spatial relationship, reduce the complexity of model processing and accurately predict the traffic speed.
Description
Technical Field
The invention relates to the field of traffic data processing, in particular to a traffic speed prediction method, a traffic speed prediction system and a storage medium.
Background
In recent years, with rapid development of various positioning technologies such as an intelligent transportation system and a Global Positioning System (GPS), a mobile device, and the like, availability of traffic data is increasing. Mining valuable knowledge from traffic data is crucial for many real-world applications, including intelligent transportation, city planning, public safety, etc. Traffic prediction has great significance for realizing traffic guidance, travel planning and congestion control. It has become more and more interesting to make real-time and accurate traffic predictions.
A traffic prediction problem is defined as predicting traffic information for a certain time period in the future based on historical traffic information in the road network. The invention focuses on traffic speed information, the traffic speed data is space-time data, and the traffic speed data has complex time and space correlation due to the constraint of the topological structure of the urban road network and the rule of dynamic change. How to accurately and comprehensively mine the space-time correlation among the traffic speed data is a key point for improving the accuracy of traffic prediction.
At present, many researches on traffic prediction methods exist at home and abroad, and the researches can be roughly divided into two types: a conventional machine learning method and a deep learning method.
The main methods of the traditional machine learning method are autoregressive integrated moving average model ARIMA [1], linear regression model [2], and a series of ARIMA model variants proposed for improving prediction accuracy, periodic ARIMA [3], subset ARIMA [4 ]. These time series models use the observed time series to predict future data. However, these models rely on the assumption that the system model is static, and cannot reflect the non-linearity and uncertainty of traffic data, and cannot overcome the interference of random events such as traffic accidents. Therefore, researchers begin to use methods such as a support vector machine regression model [5], a Bayesian network model [6] and a K neighbor model [7], and the methods can automatically use enough historical data to learn the time change rule of traffic information and overcome the assumption that only a static system is supported.
In recent years, with the rapid development of deep learning, a deep neural network model has attracted attention because it can capture the dynamic characteristics of traffic data well. Such as convolutional neural network model (CNN) [13 ]. In image processing, CNN exhibits a strong ability to model the similarity between pixels, which can also be considered as a spatial relationship. Inspired by this point, some researchers began using Convolutional Neural Networks (CNNs) to capture the neighborhood between traffic networks, while using Recurrent Neural Networks (RNNs) and variant long-term memory networks (LSTM) of RNNs, Gated Recursive Units (GRU) [14], on a temporal axis to extract temporal features. Zhang et al [8] proposed a deep learning model called ST-ResNet, which designs a residual convolution network for the three attributes based on the time proximity, periodicity and tendency, respectively, and then dynamically integrates the three networks and external factors to predict the urban pedestrian flow number. Wang et al [9] model the traffic information as a spatio-temporal matrix, predict road traffic speed and congestion sources in combination with CNN and RNN, and add an error feedback mechanism to model the emergencies such as early and late peaks and traffic accidents, thereby improving the accuracy of road traffic speed prediction. Because roads are easier to generate a representation of a graph, researchers have begun to focus on graph convolutional neural network models (GCNs) [15 ]. The T-GCN model proposed by ZHao et al [10] takes each road section as a node, generates a road network graph according to whether the road sections are connected to form an edge, and captures and obtains spatial characteristics and temporal characteristics respectively by using a graph convolution network GCN and a gated recursion unit model GRU to generate a traffic prediction result. Yu et al [11] propose an STGCN model, which uses each observation point as a node, uses the distance between two points as an edge to generate a road network graph, and uses two space-time convolution blocks to process the graph-structured sequential traffic data, thereby finally performing road network-level traffic speed prediction. Geng et al [12] developed an ST-MGCN model to make a prediction of net appointment demand. The article uses a graph to model three spatial correlations between regions separately. And after the time sequences are processed on the three graphs respectively, extracting spatial features by using a graph convolution neural network, and finally fusing the features to finally generate a demand prediction result.
The existing method has some problems in traffic prediction. When the traditional machine learning method is used for prediction, the dynamic change of traffic information along with time is considered, but the change of a topological structure on the space is ignored, so that the traffic state cannot be accurately predicted. While the deep learning method such as the CNN model is effective in modeling spatial correlation, the model interpretability of the method to the network topology relationship is poor due to the limitation that the method is only suitable for euclidean spatial data. The development of a graph convolution neural network (GCN) model provides a good solution for extracting the spatial relationship of traffic data. When the existing graph convolution neural network model is used for traffic speed prediction, the mining of the spatial features of traffic speed data is not comprehensive, only single spatial relations of adjacent road sections or adjacent regions are mostly considered, and actually, the spatial relations exist in many ways, such as road sections which are far away from each other on a map and have very similar traffic demand patterns or congestion patterns. Or the existing ST-MGCN model excavates a plurality of spatial information, but the problems of high complexity, more parameters, slow convergence and the like of the prediction model exist.
[1]M.S.Ahmed and A.R.Cook,“Analysis of freeway traffic time-seriesdata by using Box-Jenkins techniques,”Transp.Res.Rec.,no.722,pp.1–9,1979.
[2]Dudek,Grzegorz.Pattern-based local linear regression models forshort-term load forecasting[J].Electric power systems research,2016,130(JAN.):139-147.
[3]Williams B M,Hoel L A.Modeling and Forecasting Vehicular TrafficFlow asa Seasonal ARIMA Process:Theoretical Basis and Empirical Results[J].Journal of Transportation Engineering,2003,129(6):p.664-672.
[4]Lee S,Fambro D,Lee S,et al.Application of Subset AutoregressiveIntegrated Moving Average Model for Short-Term Freeway Traffic VolumeForecasting[J].Transportation Research Record Journal of the TransportationResearch Board,1999,1678(1):179-188.
[5]Wu C H,Wei C C,Su D C,et al.Travel time prediction with supportvector regression[C]//Intelligent Transportation Systems,IEEE.IEEE,2003.
[6]Sun S,Zhang C,Yu G.A Bayesian Network Approach to Traffic FlowForecasting[J].IEEE Transactions on Intelligent Transportation Systems,2006,7(1):p.124-132.
[7]ZHANG Xiao-li,HE Guo-guang,LU Hua-pu.Short-term traffic flowforecasting based on K-nearest neighbors non-parametric regression[J].journalof systems engineering,2009.
[8]Zhang J,Zheng Y,Qi D.Deep Spatio-Temporal Residual Networks forCitywide Crowd Flows Prediction[J].2016.
[9]Wang J,Gu Q,Wu J,et al.Traffic Speed Prediction and CongestionSource Exploration:A Deep Learning Method[C]//IEEE International Conferenceon Data Mining.IEEE,2016.
[10] ZHao, Ling, Song, Yujiao, Zhang, Chao, et al T-GCN A Temporal graphic relational network for Traffic Prediction [ J ].2018.
[11]Yu,Bing,Haoteng Yin,and Zhanxing Zhu."Spatio-temporal graphconvolutional networks:A deep learning framework for traffic forecasting."arXiv preprint arXiv:1709.04875(2017).
[12]Geng,Xu,et al."Spatiotemporal multi-graph convolution network forride-hailing demand forecasting."2019AAAI Conference on ArtificialIntelligence(AAAI’19).2019.
[13]Krizhevsky A,Sutskever I,Hinton G E.Imagenet classification withdeep convolutional neural networks[C].Advances in neural informationprocessing systems.2012:1097-1105.
[14]R.Fu,Z.Zhang,and L.Li,“Using LSTM and GRU neural network methodsfor traffic flow prediction,”in Proc.31st Youth AcademicAnnu.Conf.Chin.Assoc.Automat.(YAC),Wuhan,China,Nov.2016,pp.324–328.
[15]Defferrard,M.；Bresson,X.；and Vandergheynst,P.2016.Convolutionalneural networks on graphs with fast localized spectral filtering.In Advancesin Neural Information Processing Systems,3844–3852.
[16]LI,Yaguang,et al.Diffusion convolutional recurrent neuralnetwork:Data-driven traffic forecasting.arXiv preprint arXiv:1707.01926,2017.
Disclosure of Invention
The invention aims to solve the technical problem that aiming at the defects of the prior art, the invention provides a traffic speed prediction method, a traffic speed prediction system and a storage medium, which can fully and comprehensively excavate the spatial relationship, reduce the complexity of model processing and accurately predict the traffic speed.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows: a traffic speed prediction method, characterized by comprising the steps of:
1) collecting an original traffic speed data set, dividing the data set into a training set and a test set, mining different spatial relationships of the original traffic speed data set, and constructing two road network graphs;
2) fusing the adjacency matrixes of the two road network graphs into a new graph adjacency matrix;
3) taking the training set and the new image adjacency matrix as the input of a traffic prediction model, and training to obtain a prediction model;
4) and predicting the traffic speed by using the prediction model.
The invention models the road network into a graph form, can abstract the complex topological structure between roads compared with a method using a convolution neural network, constructs two different road networks to respectively represent two spatial relations in order to fully and comprehensively mine the spatial relation, fuses the adjacent matrixes of the road networks to generate a new image adjacent matrix, and the new image adjacent matrix can represent the complex spatial relation in the road network, thereby reducing the complexity of processing multiple images.
In the step 1), the two road network graphs are respectively a neighbor graph and a correlation graph; the neighbor graph GN＝(VN,EN,AN) Wherein V isNBeing nodes of an adjacency graph, ENRepresenting an edge of an adjacent graph, ANAn adjacency matrix that is a neighbor graph; the correlation graph Gs is (Vs, Es, As), wherein Vs is a node of the correlation graphAnd Es represents the side of the correlation diagram, and As is the adjacency matrix of the correlation diagram. According to the method, two spatial relations are extracted, the neighbor graph represents adjacent nodes which are close and mutually influence, the correlation graph represents nodes which are not adjacent but can still mutually influence, and compared with other methods adopting a single graph, the spatial relation is more comprehensively mined and more effectively predicted.
The nodes of the neighbor graph are roads or sensors for collecting original traffic speed data; when the node is a road,
when the node is a sensor,
wherein v isi,vjRepresentative of sensor i and sensor j, dist (v)i,vj) Represents the distance between sensor i and sensor j, σ is the standard deviation between the distances of sensor i and sensor j, and k is a threshold set to ensure the sparsity of the adjacency matrix. The invention adopts two methods for constructing the adjacency matrix of the neighbor graph, is respectively applied to the nodes which are roads or sensors, enhances the expansibility of the method, can be used for various types of real data, and can effectively model the neighbor space relationship by using the two composition methods.
The weight and adjacency matrix expressions of the edges of the dependency graph are as follows:
wherein r isi,jRepresenting nodes of the correlation graph (which may be roads or sensors collecting original traffic speed data, if the nodes of the neighbor graph are roads, the nodes of the correlation graph are roads, and if the nodes of the neighbor graph are sensors collecting original traffic speed data, the nodes of the correlation graph are original collection nodesSensor of initial traffic speed data) Pearson's coefficient, X, between i and node jiRepresenting the velocity vector of the node i,
mean value of velocity vector, Y, representing node iiA velocity vector representing the node j,
the mean of the velocity vectors representing node j; as is the adjacency matrix of the dependency graph; n is the number of nodes. The invention adopts the Pearson correlation coefficient to calculate the correlation of the velocity vector between the nodes, and the Pearson correlation coefficient is widely used for measuring the correlation degree between two variables and has the value between-1 and 1. The larger the absolute value of the correlation coefficient, the stronger the degree of linear correlation between the two vectors. Because the original traffic speed data set has higher dimensionality, compared with other methods for calculating the correlation, such as Euclidean distance measurement method and cosine similarity, the method is simpler and has stronger fault tolerance.
The specific implementation process of the step 2) comprises the following steps:
A) respectively calculating normalized Laplace matrixes of the two road network graphs;
B) respectively calculating spectrum embedding matrixes of the two road network graphs by utilizing the normalized Laplace matrixes of the two road network graphs;
C) calculating a Laplace matrix L of a new graph according to the spectrum embedding matrix and the normalized Laplace matrix of each road network graphnew＝(LN+LS)-(α1UNUN'+α2USUS') to a host; take out LnewDiagonal of (D) generates a degree matrix D of the new graphnewAccording to formula Anew＝Dnew-LnewTo find an adjacency matrix A of the new graphnew(ii) a Wherein L isNLs are normalized laplacian matrices of the two road network maps respectively; u shapeNUs are respectively the spectrum embedding matrix, U, of two road network mapsN'，US' respectively, spectrum embedding matrix UNTranspose of Us, α1，α2Is a hyperparameter balancing the number of terms of the equation.
The invention adopts the normalized Laplace matrix of the road network diagram, and the matrix normalization is the simplified operation complexity. And finally, the Laplace matrix of the new graph is obtained by utilizing the spectrum embedding matrix and the normalized Laplace matrix of the road network graph. Compared with a method of directly utilizing two single graphs, the new graph information obtained by processing through the fusion method is richer and more effective, the complexity of calculation is reduced due to dimension reduction operation, and some unimportant and redundant information influencing a prediction result is deleted.
The traffic prediction model includes a graph convolution neural network for extracting spatial features and a gate recursion unit for extracting temporal features.
Although a deep learning method such as a CNN model is effective in modeling spatial correlation, the deep learning method has poor interpretability of modeling a topological relation of a network due to the limitation that the deep learning method is only suitable for Euclidean spatial data. The traffic speed data is large space-time data, so a traffic prediction model needs to extract spatial features and temporal features at the same time, a road network in the invention is represented by a graph form, a Graph Convolution Neural Network (GCNN) naturally becomes a very suitable choice for extracting the spatial features in the scene, a common method for extracting the temporal features comprises a Recurrent Neural Network (RNN) and a variant long-time memory network (LSTM) and a Gated Recursion Unit (GRU), the recurrent neural network is lack of extraction of long-time dependency relation compared with the long-time memory network, the long-time memory network and the gated recursion unit have more parameters and slow convergence, and the gated recursion unit is selected to extract the temporal features. Therefore, the traffic prediction model of the invention is a graph convolution neural network for extracting spatial features and a gate recursion unit for extracting temporal features.
The graph convolution neural network includes:
the input layer inputs the original traffic speed data set and the new map adjacency matrix;
the hidden layer is used for extracting spatial features in the road network graph and performing graph convolution operation on the new graph adjacency matrix, and the graph convolution operation formula is as follows:
where X is the original traffic speed data set, Anew is the new map adjacency matrix,
a to Anew + I, I is a unit matrix, D to ∑ A to D is a corresponding degree matrix of A to W0 and W1 are weight matrix from input layer to hidden layer and weight matrix from hidden layer to output layer respectively.
An output layer for outputting a result f (X, Anew) of the graph convolution operation;
wherein, the activation functions of the hidden layer and the output layer are both ReLU functions.
The gated recursion unit comprises:
the input layer inputs the output result f (X) obtained from the graph convolution neural network at the time t-1t-1Anew) and the hidden state h at time t-1t-1Wherein X ist-1Traffic speed data representing original traffic speed data set X at time t-1
A hidden layer for obtaining an output result f (X) from the graph convolution neural network according to the time t-1t-1Anew) and the hidden state h at time t-1t-1The current hidden state h is obtained by the following formulat：
ut＝σ(Wu·[ht-1,f(Xt-1,Anew)])
rt＝σ(Wr·[ht-1,f(Xt-1,Anew)])
c＝tanh(Wc·[(rt*ht-1),f(Xt-1,Anew)])；
ht＝(1-ut)*c+ut*ht-1
Where u is the update gate with the input being the output f (X) obtained from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wu is a weight matrix connected input to the update gate, utIs the output of the update gate; r is a reset gate with an input of the output f (X) from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wr is a weight matrix connecting the input layer to the reset gate, rtIs the output of the reset gate; c is the value of the candidate hidden state, input as the output result f (X) obtained from the atlas neural network at time t-1t-1Anew) and the hidden state h of the reset gate output and time t-1t-1Wc is a weight matrix connecting the input and the candidate hidden state; h istIs the current hidden state; σ () is Sigmoid function, tanh is hyperbolic tangent function;
an output layer for outputting the hidden state h at the current momentt。
The value ranges of the hidden layer nodes of the graph convolution neural network and the gated recursion unit are [16, 128], the error is ensured to be proper, and the complexity of a prediction model is reduced.
The present invention also provides a traffic speed prediction system comprising a computer device configured or programmed to perform the steps of the above method.
As an inventive concept, the present invention also provides a computer-readable storage medium storing a program for executing the steps of the above-described method.
Compared with the prior art, the invention has the beneficial effects that:
1. the invention can fully and comprehensively excavate the spatial relationship, the adjacent matrix of the new graph can represent the complex spatial relationship in the road network, the complexity of processing multiple graphs is reduced, and the traffic speed can be accurately predicted;
2. compared with the prior art, the prediction model is simple, the calculated amount in the training process is small, the convergence is fast, and the practicability is strong.
Drawings
FIG. 1 is a frame diagram of a traffic multi-map prediction model FMGCN proposed by the present invention;
FIG. 2 is a schematic diagram of a two-layer graph convolutional neural network employed by the present invention;
FIG. 3 is an internal block diagram of a gated recursion unit for use with the present invention;
FIG. 4 is an internal structure diagram of a space-time graph convolution prediction model (TGCN);
FIG. 5 is a network architecture diagram of a multi-graph based network appointment demand prediction model (ST-MGCN);
FIG. 6 is a bar graph of the comparison of the accuracies of the 5 models;
FIG. 7 is a bar graph of accuracy comparison using a single graph for the proposed prediction model.
Detailed Description
The embodiment of the invention uses a multi-view graph convolution neural network model to predict the traffic speed, and the method is divided into five parts, wherein the first part is data preprocessing and graph construction, the second part is graph fusion, the third part is prediction model construction, the fourth part is a training prediction model, and the fifth part is test experiment effect.
The first step, data preprocessing, including two aspects, firstly, preprocessing the original traffic speed data, and secondly, generating different spatial relationship diagrams.
1. Raw traffic speed data set. The originally collected traffic speed data usually does not meet the requirements, and can meet the requirements after being processed. The first step of the processing is to perform data normalization, and the second step is to construct a training set and a test set.
The purpose of normalization is to alleviate the problem that the neural network training time is increased and an optimal solution may not be found due to too large differences in the data samples. The purpose of training the neural network is to find the optimal solution of the objective function, if the sample characteristics X1 and X2 are [1,10000] and [1,100], respectively, the model takes both effects into consideration, and the difference between data samples will cause the path for finding the optimal solution to be lengthened, thereby lengthening the training time and possibly falling into local optimization. After the data is normalized, the defect can be eliminated.
The invention adopts a linear normalization method, and the conversion formula is as follows:
where X is the original feature value, X' is the normalized value of X, min (X) is the minimum value of X in the original data set, and max (X) is the maximum value of X in the data set.
The normalized data set is then divided into a training set and a test set. The samples in the training set are used for the model to learn the characteristic of traffic speed change, and the samples in the testing set are used for verifying the effect of the model. The present invention uses a leave-out method to divide the normalized data set into mutually exclusive training sets and test sets, i.e., the first 80% of the samples constitute the training set, and the remaining 20% of the samples serve as the test set.
The invention can also adopt methods such as cross-validation method, self-service method and the like to divide the training set and the test set, and the set-out method (hold-out) is to directly divide the data set D into two mutually exclusive sets, wherein one set is used as the training set S, and the other set is used as the test set T. After training the model at S, T is used to estimate its test error as an estimate of the generalization error. The cross-validation method is that every sample data is used as both training data and test data. When the data set is large, for deep learning, if the number of training samples is M, the calculation overhead for training M models is too large to be tolerated. The bootstrap method is based on a bootstrap sampling method, so the resulting data set changes the distribution of the initial data set, and estimation bias is introduced. Therefore, when the initial data amount is sufficient, the leave-out method and the cross-validation method are more commonly used, and in order to reduce the overhead, the leave-out method is adopted in the invention to divide the training set and the test set.
2. And (5) constructing a graph. The invention excavates two spatial relations to form two different graphs and generates two adjacent matrixes. Two spatial relationship maps are generated in the following manner.
Neighbor graph, GN＝(VN,EN,AN)。VNFor nodes of the adjacency graph, a road or a sensor originally acquiring traffic speed can be regarded as a node, ENRepresenting adjacent picturesEdge, ANIs an adjacency matrix of a neighbor graph. If a road is considered as a node, the elements of the adjacency matrix are only 0, 1.
If the sensor is considered as a node, the calculation formula of the adjacency matrix is as follows:
wherein v isi,vjRepresentative of sensor i and sensor j, dist (v)i,vj) Represents the distance between sensor i and sensor j, σ is the standard deviation between the distances of the sensors, and k is a threshold set to ensure the sparsity of the matrix, which is set to 0.1 in the present invention.
The correlation plot, Gs ═ Vs, Es, As. And evaluating the spatial dependence between the two nodes by utilizing the correlation of the historical speed observed value sequence between the nodes. We use the pearson correlation coefficient to compute the similarity between any two nodes as weights (the values of the elements of the adjacency matrix) for the edges of the correlation graph. The calculation formula of the pearson correlation coefficient is as follows:
wherein r isi,jRepresenting the Pearson coefficient, X, between node i and node jiRepresenting the velocity vector of the node i,
mean value of velocity vector, Y, representing node iiA velocity vector representing the node j,
represents the mean of the velocity vectors for node j and n represents the number of nodes.
And secondly, fusing the graphs. After the data is processed, the further work is to generate a new graph.
1. For two single graphs (road network graph) G generated in the first stepNGs all calculate a normalized Laplace matrix LNFor an undirected graph G (V, E, W), if D is defined as the degree matrix of the graph and W is defined as the adjacency matrix of the graph, the normalized laplace matrix of the graph is defined as:
2, calculating two simple graphs GNGs spectral embedding matrix UNUs, the spectrum embedding matrix respectively includes LNAnd the k minimum eigenvalues of Ls correspond to eigenvectors, and in the embodiment, the value of k is 2.
3, calculating a Laplace matrix L of a new graph according to the spectrum embedding matrix and the normalized Laplace matrix of each road network graphnew＝(LN+LS)-(α1UNUN'+α2USUS') to a host; take out LnewDiagonal of (D) generates a degree matrix D of the new graphnewAccording to formula Anew＝Dnew-LnewTo find an adjacency matrix A of the new graphnew(ii) a Wherein L isNLs are normalized laplacian matrices of the two road network maps respectively; u shapeNUs are respectively the spectrum embedding matrix, U, of two road network mapsN'，US' respectively, spectrum embedding matrix UNTranspose of Us, α1，α2To balance the over-parameters of the terms of the equation, for convenience of calculation, α is ordered in the present invention1＝α20.5. And taking out a diagonal line of the Lnew to generate a degree matrix Dnew of the new graph, and solving an adjacent matrix Anew of the new graph according to a formula Anew which is Dnew-Lnew.
And thirdly, constructing a traffic prediction model.
Traffic-based prediction is a spatio-temporal prediction problem, so two separate models are constructed to deal with spatial and temporal features separately. The specific structure is shown in figure 1, in the invention, a graph convolution neural network is adopted to extract spatial features, and a gated recursion unit is adopted to extract temporal features.
A historical time window and a predicted time window are first determined. The size of the historical time window represents how many time instants the traffic speed value was used at in the past. The size of the prediction time window represents how many moments the traffic speed value in the next time is predicted.
Secondly, the number of neurons of the hidden layer is determined. In particular, the number of hidden layer units of the graph convolution neural network and the gated recursion unit needs to be determined respectively.
When designing a neural network, finding an appropriate number of nodes of a hidden layer plays a significant role in the performance of a neural network model. The overlarge hidden layer is also one of the reasons for the occurrence of the overfitting phenomenon, and in order to prevent the overfitting situation as much as possible and to make the performance effect of the prediction model better, the most basic principle adopted by the invention is as follows: the reasonable hidden layer node number considers both the error size and the complexity of the prediction model, and the invention sets the value range of the hidden layer node number as [16, 128], such as 16,32,64,128 and the like. The specific size of the hidden layer node is given in the experimental analysis section.
Finally, an output layer is constructed. The number of neurons in the output layer depends on the prediction time window. In the invention, the output layer only has one neuron and receives the hidden layer input at the last moment.
After the input layer, the hidden layer and the output layer are determined, the layers are fully connected with weights, and the model is built.
Specifically, the design of this embodiment adopts a two-layer graph convolution neural network, the specific structure is shown in fig. 2, and the forward propagation model of the two-layer graph convolution neural network is
The structure of the gated recursion unit is shown in FIG. 3, which is expressed as
ut＝σ(Wu·[ht-1,f(Xt-1,Anew)])
rt＝σ(Wr·[ht-1,f(Xt-1,Anew)])
c＝tanh(Wc·[(rt*ht-1),f(Xt-1,Anew)])
ht＝(1-ut)*c+ut*ht-1
Where u is the update gate with the input being the output f (X) obtained from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wu is a weight matrix connected input to the update gate, utIs the output of the update gate; r is a reset gate with an input of the output f (X) from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wr is a weight matrix connecting the input layer to the reset gate, rtIs the output of the reset gate; c is the value of the candidate hidden state, input as the output result f (X) obtained from the atlas neural network at time t-1t-1Anew) and the hidden state h of the reset gate output and time t-1t-1Wc is a weight matrix connecting the input and the candidate hidden state; h istIs the current hidden state; σ () is Sigmoid function, tanh is hyperbolic tangent function
And fourthly, training a traffic prediction model.
The training process of the neural network is an optimization loss function process. The loss function (the loss of the training set is not reduced, and the convergence state is considered to be achieved) in the invention uses a 2-norm loss function, in order to avoid overfitting, a 2-norm regularization term is added, and the loss function is recorded as:
Loss＝||Ypred-Ytrue||+l2where Ypred represents the predicted value and Ytrue represents the true value.
The present invention employs a back propagation algorithm (BP algorithm) based training neural network. The BP algorithm is established on the basis of a gradient descent method, and is a learning algorithm suitable for a multilayer neuron network. And an Adam optimizer is employed to promote the algorithmic model.
The fifth step is to check the experimental effect
After the model is trained, the effect of the model is checked on a test set, test samples are input into the model one by one, and the prediction accuracy of the model on the test set is calculated. And corresponding evaluation indexes are needed for evaluating the effect of the prediction model. The indicators taken by the present invention are Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). Assuming that N is the number of samples, the calculation formula of the evaluation index is as follows:
the smaller the MAE and RMSE values, the better the model performance.
The performance of the method of the invention was analyzed experimentally as follows:
in order to prove that the traffic prediction method provided by the invention has better effect than the existing traffic prediction method, the several compared models in the experiment are respectively as follows: a historical average model HA, a support vector machine regression model SVR, a T-GCN model and a ST-MGCN model. HA and SVR are traditional machine learning methods, T-GCN is a single graph model structure as shown in figure 4, and ST-MGCN is a multi-graph model structure as shown in figure 5.
The data set used in the experiment was the METR-LA data set provided in the paper DCRNN [16], which was collected from a circular detector on the los Angeles highway, including four months of traffic speed data from 3/1/2012 to 6/30/6, collected every 5 min. The method adopts 207 sensors, traffic speed data from 3/1/2012 to 3/7/2012, 2016 data points are obtained, input data are normalized to a [0,1] interval, the first 80% of data are used as a training set, and the last 20% of data are used as a test set by a leave-out method. The historical time window is set to 12 and the predicted time window is set to 3, that is, the historical one hour traffic speed is used to predict the traffic speed 15 minutes in the future. In order to make the model as convergent as possible, the learning rate was set to 0.001, and after testing, the hidden layer neural unit was found to be set to 64 to be optimal. The input samples are converted into a three-dimensional tensor input, [ batch _ size, seq _ len, nodes ], where batch _ size represents the number of samples that need to be input to update the weights once, seq _ len is the length of the input sequence, and nodes represents the number of nodes. Their specific values of 3 are 32, 12, and 207, respectively.
FIG. 6 is a bar graph of the comparison of the accuracies of the 5 models. The following experimental results are generated by iterating the model for 50 times, and as can be seen from the figure, the prediction model provided by the invention is superior to other four methods in the aspect of RMSE, but the prediction model provided by the invention is not greatly different from the SVR results in the aspect of MAE, but is superior to other three models.
FIG. 7 is a bar graph of accuracy comparison using a single graph for the proposed prediction model. In order to verify the effect of the invention, the adjacency graph and the correlation graph are used for respectively predicting, and the result shows that the single graph effect generated by the multi-graph fusion method adopted by the invention is optimal, which shows that the method provided by the invention effectively excavates various spatial relationships, and can effectively integrate various spatial relationships on the premise of reducing the complexity of the model, thereby improving the prediction precision.
Claims (10)
1. A traffic speed prediction method, characterized by comprising the steps of:
1) collecting an original traffic speed data set, dividing the data set into a training set and a test set, mining different spatial relationships of the original traffic speed data set, and constructing two road network graphs;
2) fusing the adjacency matrixes of the two road network graphs into a new graph adjacency matrix;
3) taking the training set and the new image adjacency matrix as the input of a traffic prediction model, and training to obtain a prediction model;
4) and predicting the traffic speed by using the prediction model.
2. The traffic speed prediction method according to claim 1, wherein in step 1), the two road network maps are a neighbor map and a correlation map respectively; the neighbor graph GN＝(VN,EN,AN) Wherein V isNBeing nodes of an adjacency graph, ENRepresenting an edge of an adjacent graph, ANAn adjacency matrix that is a neighbor graph; and the correlation map Gs is (Vs, Es, As), wherein Vs is a node of the correlation map, Es represents an edge of the correlation map, and As is an adjacent matrix of the correlation map.
3. The traffic speed prediction method according to claim 2, wherein the nodes of the neighborhood graph are roads or sensors that collect raw traffic speed data; when the node is a road,
when the node is a sensor,
wherein v isi,vjRepresentative of sensor i and sensor j, dist (v)i,vj) Represents the distance between sensor i and sensor j, σ is the standard deviation between the distances of sensor i and sensor j, and k is a threshold set to ensure the sparsity of the adjacency matrix.
4. The traffic speed prediction method according to claim 2, wherein the weight of the edge of the correlation graph and the adjacency matrix expression are as follows:
wherein r isi,jRepresenting the Pearson coefficient, X, between nodes i and j of the correlation graphiRepresenting the velocity vector of the node i,
mean value of velocity vector, Y, representing node iiA velocity vector representing the node j,
the mean of the velocity vectors representing node j; as is the adjacency matrix of the dependency graph; n is the number of nodes.
5. The traffic speed prediction method according to claim 1, wherein the concrete implementation process of step 2) includes:
A) respectively calculating normalized Laplace matrixes of the two road network graphs;
B) respectively calculating spectrum embedding matrixes of the two road network graphs by utilizing the normalized Laplace matrixes of the two road network graphs;
C) calculating a Laplace matrix L of a new graph according to the spectrum embedding matrix and the normalized Laplace matrix of each road network graphnew＝(LN+LS)-(α1UNUN'+α2USUS') (ii) a Take out LnewDiagonal of (D) generates a degree matrix D of the new graphnewAccording to formula Anew＝Dnew-LnewTo find an adjacency matrix A of the new graphnew(ii) a Wherein L isNLs are normalized laplacian matrices of the two road network maps respectively; u shapeNUs are respectively the spectrum embedding matrix, U, of two road network mapsN'，US'Respectively, spectrum embedding matrix UNTranspose of Us, α1，α2Is a hyperparameter balancing the number of terms of the equation.
6. The traffic speed prediction method according to claim 1, characterized in that the traffic prediction model comprises a graph convolution neural network for extracting spatial features and a gated recursion unit for extracting temporal features.
7. The traffic speed prediction method of claim 6, wherein the graph convolutional neural network comprises:
the input layer inputs the original traffic speed data set and the new map adjacency matrix;
the hidden layer is used for extracting spatial features in the road network graph and performing graph convolution operation on the new graph adjacency matrix, and the graph convolution operation formula is as follows:
where X is the original traffic speed data set, Anew is the new map adjacency matrix,
a to Anew + I, I is a unit matrix, D to ∑ A to D is a corresponding degree matrix of A to W0 and W1 are a weight matrix from an input layer to a hidden layer and a weight matrix from the hidden layer to an output layer respectively;
an output layer for outputting a result f (X, Anew) of the graph convolution operation; f (X, Anew), i.e., spatial features;
wherein, the activation functions of the hidden layer and the output layer are both ReLU functions;
preferably, the number of hidden layer nodes ranges from [16, 128 ].
8. The traffic speed prediction method according to claim 7, wherein the gated recursion unit comprises:
the input layer inputs the output result f (X) obtained from the graph convolution neural network at the time t-1t-1Anew) and t-1Hidden state of carving ht-1Wherein X ist-1Traffic speed data representing an original traffic data set X at time t-1;
a hidden layer for obtaining an output result f (X) from the graph convolution neural network according to the time t-1t-1Anew) and the hidden state h at time t-1t-1The current hidden state h is obtained by the following formulat：
ut＝σ(Wu·[ht-1,f(Xt-1,Anew)])
rt＝σ(Wr·[ht-1,f(Xt-1,Anew)])
c＝tanh(Wc·[(rt*ht-1),f(Xt-1,Anew)])；
ht＝(1-ut)*c+ut*ht-1
Where u is the update gate with the input being the output f (X) obtained from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wherein X ist-1Traffic speed data representing the original traffic data set X at time t-1, Wu being a weight matrix connected input to an update gate, utIs the output of the update gate; r is a reset gate with an input of the output f (X) from the convolutional neural network at time t-1t-1Anew) and the hidden state h at time t-1t-1Wr is a weight matrix connecting the input layer to the reset gate, rtIs the output of the reset gate; c is the value of the candidate hidden state, input as the output result f (X) obtained from the atlas neural network at time t-1t-1Anew) and the hidden state h of the reset gate output and time t-1t-1Wc is a weight matrix connecting the input layer and the candidate hidden state; h istIs the current hidden state; σ () is Sigmoid function, tanh is hyperbolic tangent function;
an output layer for outputting the hidden state h at the current momentt；htI.e., a temporal feature;
preferably, the number of hidden layer nodes ranges from [16, 128 ].
9. A traffic speed prediction system comprising a computer device, characterized in that the computer device is configured or programmed for performing the steps of the method according to any one of claims 1 to 8.
10. A computer-readable storage medium storing a program for performing the steps of the method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010418444.2A CN111612243B (en) | 2020-05-18 | 2020-05-18 | Traffic speed prediction method, system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010418444.2A CN111612243B (en) | 2020-05-18 | 2020-05-18 | Traffic speed prediction method, system and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111612243A true CN111612243A (en) | 2020-09-01 |
| CN111612243B CN111612243B (en) | 2023-11-10 |
Family
ID=72201878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010418444.2A Active CN111612243B (en) | 2020-05-18 | 2020-05-18 | Traffic speed prediction method, system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111612243B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112201346A (en) * | 2020-10-12 | 2021-01-08 | 哈尔滨工业大学（深圳） | Cancer survival prediction method, apparatus, computing device and computer-readable storage medium |
| CN112216108A (en) * | 2020-10-12 | 2021-01-12 | 中南大学 | Traffic prediction method based on attribute-enhanced space-time graph convolution model |
| CN112241814A (en) * | 2020-10-20 | 2021-01-19 | 河南大学 | Traffic prediction method based on reinforced space-time diagram neural network |
| CN112257614A (en) * | 2020-10-26 | 2021-01-22 | 中国民航大学 | Station building passenger flow space-time distribution prediction method based on graph convolution network |
| CN112562312A (en) * | 2020-10-21 | 2021-03-26 | 浙江工业大学 | GraphSAGE traffic network data prediction method based on fusion characteristics |
| CN112712695A (en) * | 2020-12-30 | 2021-04-27 | 桂林电子科技大学 | Traffic flow prediction method, device and storage medium |
| CN112766551A (en) * | 2021-01-08 | 2021-05-07 | 鹏城实验室 | Traffic prediction method, intelligent terminal and computer readable storage medium |
| CN112863180A (en) * | 2021-01-11 | 2021-05-28 | 腾讯大地通途（北京）科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
| CN112991721A (en) * | 2021-02-04 | 2021-06-18 | 南通大学 | Urban road network traffic speed prediction method based on graph convolution network node association degree |
| CN113112819A (en) * | 2021-03-26 | 2021-07-13 | 华南理工大学 | Improved LSTM-based graph convolution traffic speed prediction method |
| CN113160570A (en) * | 2021-05-27 | 2021-07-23 | 长春理工大学 | Traffic jam prediction method and system |
| CN113268916A (en) * | 2021-04-07 | 2021-08-17 | 浙江工业大学 | Traffic accident prediction method based on space-time graph convolutional network |
| CN113393017A (en) * | 2021-05-07 | 2021-09-14 | 浙江警官职业学院 | Individual complex relationship-based supervision place safety early warning method and device |
| CN113611119A (en) * | 2021-08-10 | 2021-11-05 | 北京航空航天大学 | Vehicle induction method based on gated recursion unit |
| CN116402554A (en) * | 2023-06-07 | 2023-07-07 | 江西时刻互动科技股份有限公司 | Advertisement click rate prediction method, system, computer and readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109685252A (en) * | 2018-11-30 | 2019-04-26 | 西安工程大学 | Building energy consumption prediction technique based on Recognition with Recurrent Neural Network and multi-task learning model |
| CN109754126A (en) * | 2019-01-30 | 2019-05-14 | 银江股份有限公司 | Short-time Traffic Flow Forecasting Methods based on temporal correlation and convolutional neural networks |
| CN110517479A (en) * | 2018-05-22 | 2019-11-29 | 杭州海康威视***技术有限公司 | A kind of urban highway traffic prediction technique, device and electronic equipment |
| CN110969854A (en) * | 2019-12-13 | 2020-04-07 | 深圳先进技术研究院 | Traffic flow prediction method, system and terminal equipment |
| CN111079975A (en) * | 2019-11-14 | 2020-04-28 | 青岛海信网络科技股份有限公司 | Traffic data prediction method and device and vehicle control method |
-
2020
- 2020-05-18 CN CN202010418444.2A patent/CN111612243B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110517479A (en) * | 2018-05-22 | 2019-11-29 | 杭州海康威视***技术有限公司 | A kind of urban highway traffic prediction technique, device and electronic equipment |
| CN109685252A (en) * | 2018-11-30 | 2019-04-26 | 西安工程大学 | Building energy consumption prediction technique based on Recognition with Recurrent Neural Network and multi-task learning model |
| CN109754126A (en) * | 2019-01-30 | 2019-05-14 | 银江股份有限公司 | Short-time Traffic Flow Forecasting Methods based on temporal correlation and convolutional neural networks |
| CN111079975A (en) * | 2019-11-14 | 2020-04-28 | 青岛海信网络科技股份有限公司 | Traffic data prediction method and device and vehicle control method |
| CN110969854A (en) * | 2019-12-13 | 2020-04-07 | 深圳先进技术研究院 | Traffic flow prediction method, system and terminal equipment |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112216108A (en) * | 2020-10-12 | 2021-01-12 | 中南大学 | Traffic prediction method based on attribute-enhanced space-time graph convolution model |
| CN112201346A (en) * | 2020-10-12 | 2021-01-08 | 哈尔滨工业大学（深圳） | Cancer survival prediction method, apparatus, computing device and computer-readable storage medium |
| CN112201346B (en) * | 2020-10-12 | 2024-05-07 | 哈尔滨工业大学（深圳） | Cancer lifetime prediction method, device, computing equipment and computer readable storage medium |
| CN112216108B (en) * | 2020-10-12 | 2021-06-29 | 中南大学 | Traffic prediction method based on attribute-enhanced space-time graph convolution model |
| CN112241814A (en) * | 2020-10-20 | 2021-01-19 | 河南大学 | Traffic prediction method based on reinforced space-time diagram neural network |
| CN112562312A (en) * | 2020-10-21 | 2021-03-26 | 浙江工业大学 | GraphSAGE traffic network data prediction method based on fusion characteristics |
| CN112257614A (en) * | 2020-10-26 | 2021-01-22 | 中国民航大学 | Station building passenger flow space-time distribution prediction method based on graph convolution network |
| CN112257614B (en) * | 2020-10-26 | 2022-05-17 | 中国民航大学 | Station building passenger flow space-time distribution prediction method based on graph convolution network |
| CN112712695B (en) * | 2020-12-30 | 2021-11-26 | 桂林电子科技大学 | Traffic flow prediction method, device and storage medium |
| CN112712695A (en) * | 2020-12-30 | 2021-04-27 | 桂林电子科技大学 | Traffic flow prediction method, device and storage medium |
| CN112766551A (en) * | 2021-01-08 | 2021-05-07 | 鹏城实验室 | Traffic prediction method, intelligent terminal and computer readable storage medium |
| CN112766551B (en) * | 2021-01-08 | 2022-05-17 | 鹏城实验室 | Traffic prediction method, intelligent terminal and computer readable storage medium |
| CN112863180A (en) * | 2021-01-11 | 2021-05-28 | 腾讯大地通途（北京）科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
| CN112863180B (en) * | 2021-01-11 | 2022-05-06 | 腾讯大地通途（北京）科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
| CN112991721A (en) * | 2021-02-04 | 2021-06-18 | 南通大学 | Urban road network traffic speed prediction method based on graph convolution network node association degree |
| CN113112819A (en) * | 2021-03-26 | 2021-07-13 | 华南理工大学 | Improved LSTM-based graph convolution traffic speed prediction method |
| CN113268916A (en) * | 2021-04-07 | 2021-08-17 | 浙江工业大学 | Traffic accident prediction method based on space-time graph convolutional network |
| CN113393017B (en) * | 2021-05-07 | 2022-08-23 | 浙江警官职业学院 | Individual complex relationship-based supervision place safety early warning method and device |
| CN113393017A (en) * | 2021-05-07 | 2021-09-14 | 浙江警官职业学院 | Individual complex relationship-based supervision place safety early warning method and device |
| CN113160570A (en) * | 2021-05-27 | 2021-07-23 | 长春理工大学 | Traffic jam prediction method and system |
| CN113611119A (en) * | 2021-08-10 | 2021-11-05 | 北京航空航天大学 | Vehicle induction method based on gated recursion unit |
| CN113611119B (en) * | 2021-08-10 | 2022-06-28 | 北京航空航天大学 | Vehicle induction method based on gated recursion unit |
| CN116402554A (en) * | 2023-06-07 | 2023-07-07 | 江西时刻互动科技股份有限公司 | Advertisement click rate prediction method, system, computer and readable storage medium |
| CN116402554B (en) * | 2023-06-07 | 2023-08-11 | 江西时刻互动科技股份有限公司 | Advertisement click rate prediction method, system, computer and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111612243B (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111612243B (en) | Traffic speed prediction method, system and storage medium | |
| CN109754605B (en) | Traffic prediction method based on attention temporal graph convolution network | |
| CN110223517B (en) | Short-term traffic flow prediction method based on space-time correlation | |
| CN109492822B (en) | Air pollutant concentration time-space domain correlation prediction method | |
| Zhang et al. | DNEAT: A novel dynamic node-edge attention network for origin-destination demand prediction | |
| CN110766942B (en) | Traffic network congestion prediction method based on convolution long-term and short-term memory network | |
| CN111383452A (en) | Method for estimating and predicting short-term traffic running state of urban road network | |
| CN112949828B (en) | Graph convolution neural network traffic prediction method and system based on graph learning | |
| Ge et al. | Temporal graph convolutional networks for traffic speed prediction considering external factors | |
| CN114299723B (en) | Traffic flow prediction method | |
| CN111047078B (en) | Traffic characteristic prediction method, system and storage medium | |
| CN110570035B (en) | People flow prediction system for simultaneously modeling space-time dependency and daily flow dependency | |
| Zheng et al. | Hybrid deep learning models for traffic prediction in large-scale road networks | |
| CN113112791A (en) | Traffic flow prediction method based on sliding window long-and-short term memory network | |
| CN115376317B (en) | Traffic flow prediction method based on dynamic graph convolution and time sequence convolution network | |
| CN115206092A (en) | Traffic prediction method of BiLSTM and LightGBM model based on attention mechanism | |
| CN112766603A (en) | Traffic flow prediction method, system, computer device and storage medium | |
| CN115759461A (en) | Internet of things-oriented multivariate time sequence prediction method and system | |
| Chen et al. | A short-term traffic flow prediction model based on AutoEncoder and GRU | |
| CN115936069A (en) | Traffic flow prediction method based on space-time attention network | |
| CN112927507B (en) | Traffic flow prediction method based on LSTM-Attention | |
| CN113112792A (en) | Multi-module traffic intensity prediction method based on semantic information | |
| CN117494034A (en) | Air quality prediction method based on traffic congestion index and multi-source data fusion | |
| CN117131979A (en) | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism | |
| Chen et al. | Improved LSTM based on attention mechanism for short-term traffic flow prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |